The Core Clock Gene, Bmal1, and Its Downstream Target, the SNARE Regulatory Protein Secretagogin, Are Necessary for Circadian Secretion of Glucagon-Like Peptide-1

Original Article

The core clock gene, Bmal1, and its downstream target, the SNARE regulatory protein secretagogin, are necessary for circadian secretion of glucagon-like peptide-1

Andrew D. Biancolin 1,8, Alexandre Martchenko 1,8, Emilia Mitova 1, Patrick Gurges 1, Everan Michalchyshyn 1, Jennifer A. Chalmers 1, Alessandro Doria 2,3, Josyf C. Mychaleckyj 4, Alice E. Adriaenssens 5, Frank Reimann 5, Fiona M. Gribble 5, Manuel Gil-Lozano 1,9, Brian J. Cox 1,6, Patricia L. Brubaker 1,7,*

ABSTRACT

Objectives: The incretin hormone glucagon-like peptide-1 (GLP-1) is secreted from intestinal L-cells upon nutrient intake. While recent evidence has shown that GLP-1 is released in a circadian manner in rats, whether this occurs in mice and if this pattern is regulated by the circadian clock remain to be elucidated. Furthermore, although circadian GLP-1 secretion parallels expression of the core clock gene Bmal1, the link between the two remains largely unknown. Secretagogin (Scgn) is an exocytotic SNARE regulatory protein that demonstrates circadian expression and is essential for insulin secretion from b-cells. The objective of the current study was to establish the necessity of the core clock gene Bmal1 and the SNARE protein SCGN as essential regulators of circadian GLP-1 secretion. Methods: Oral glucose tolerance tests were conducted at different times of the day on 4-hour fasted C57BL/6J, Bmal1 wild-type, and Bmal1 knockout mice. Mass spectrometry, RNA-seq, qRT-PCR and/or microarray analyses, and immunostaining were conducted on murine (m) and human (h) primary L-cells and mGLUTag and hNCI-H716 L-cell lines. At peak and trough GLP-1 secretory time points, the mGLUTag cells were co-stained for SCGN and a membrane-marker, ChIP was used to analyze BMAL1 binding sites in the Scgn promoter, protein interaction with SCGN was tested by co-immunoprecipitation, and siRNA was used to knockdown Scgn for GLP-1 secretion assay. Results: C57BL/6J mice displayed a circadian rhythm in GLP-1 secretion that peaked at the onset of their feeding period. Rhythmic GLP-1 release was impaired in Bmal1 knockout (KO) mice as compared to wild-type controls at the peak (p < 0.05) but not at the trough secretory time point. Microarray identified SNARE and transport vesicle pathways as highly upregulated in mGLUTag L-cells at the peak time point of GLP-1 secretion (p < 0.001). Mass spectrometry revealed that SCGN was also increased at this time (p < 0.001), while RNA-seq, qRT-PCR, and immunostaining demonstrated Scgn expression in all human and murine primary L-cells and cell lines. The mGLUTag and hNCI-H716 L-cells exhibited circadian rhythms in Scgn expression (p < 0.001). The ChIP analysis demonstrated increased binding of BMAL1 only at the peak of Scgn expression (p < 0.01). Immunocytochemistry showed the translocation of SCGN to the cell membrane after stimulation at the peak time point only (p < 0.05), while CoIP showed that SCGN was pulled down with SNAP25 and b-actin, but only the latter interaction was time- dependent (p < 0.05). Finally, Scgn siRNA-treated cells demonstrated significantly blunted GLP-1 secretion (p < 0.01) in response to stimulation at the peak time point only. Conclusions: These data demonstrate, for the first time, that mice display a circadian pattern in GLP-1 secretion, which is impaired in Bmal1 knockout mice, and that Bmal1 regulation of Scgn expression plays an essential role in the circadian release of the incretin hormone GLP-1. Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Keywords Bmal1; Circadian; GLP-1; L-cell; Secretagogin; Secretion

1Department of Physiology, University of Toronto, Toronto, ON, Canada 2Department of Medicine, Harvard Medical School, Boston, MA, USA 3Research Division, Joslin Diabetes Center, Boston, MA, USA 4Center for Public Health Genomics, University of Virginia, Charlottesville, VA, USA 5Wellcome Trust-MRC Institute of Metabolic Science, University of Cambridge, Cambridge, UK 6Department of Obstetrics and Gynecology, University of Toronto, Toronto, ON, Canada 7Department of Medicine, University of Toronto, Toronto, ON, Canada

8 Indicates equal co-ﬁrst authors. 9 Currently at the Institute for Diabetes and Cancer, Helmholtz Zentrum München, Oberschleibheim, Germany.

*Corresponding author. University of Toronto, 1 King’s College Circle, Room 3366 MSB, Toronto, ON, M5S 1A8, Canada. Fax: þ1 416 978 4940. E-mail: p.brubaker@ utoronto.ca (P.L. Brubaker).

Received August 28, 2019 Revision received October 24, 2019 Accepted November 1, 2019 Available online 21 November 2019 https://doi.org/10.1016/j.molmet.2019.11.004

124 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com 1. INTRODUCTION Bmal1. We also report that Scgn is expressed in intestinal L-cells, where it exhibits circadian expression under the transcriptional regu- Circadian rhythms act as an anticipatory mechanism preparing or- lation of BMAL1. This drives a subsequent time-dependent recruitment ganisms for the constant 24-hour lightedark cycle [1,2]. The main of SCGN toward the L-cell membrane that in turn facilitates circadian zeitgeber (ZT), light, entrains a network of clock genes in the supra- secretion of GLP-1. When taken together, we identified a novel chiasmatic nuclei of the hypothalamus, where these rhythms are regulator of circadian GLP-1 secretion, which could have implications orchestrated [1,3]. At the molecular level, these rhythms are generated for time-sensitive treatments as well as the potential for SNAREs as by heterodimerization of the core clock protein BMAL1 with CLOCK and targets for type 2 diabetes therapies. subsequent binding to E-box promoter elements to stimulate the transcription of Period (Per)1e3 and Cryptochrome (Cry)1e2. The 2. METHODS expression of Per and Cry, in turn, represses Bmal1 and Clock, completing the transcriptional feedback loop. The clock genes are 2.1. Animals þ known to be expressed in all nucleated mammalian cells, both cen- Male and female C57Bl/6J mice and Bmal1 / mice were purchased þ trally and in peripheral tissues, and are estimated to drive the rhythmic from Jackson Laboratories (Bar Harbor, ME, USA). The Bmal1 / mice expression of approximately 43% of all protein-encoding genes [1,4]. were bred and genotyped according to the recommended protocol to Although light is the strongest ZT, nutrient intake can also synchronize generate sex-, age-, and littermate-matched wild-type (WT) and KO peripheral metabolic tissues. The gastrointestinal tract, b-cell, liver, animals. The mice had free access to water and a regular chow diet skeletal muscle, and adipose tissue [5e14] have been shown to (Teklad) for the duration of the study and were allowed to acclimate for exhibit endogenous circadian rhythmicity, ultimately coordinating one week to the 12-hour light, 12-hour dark (lights on at 06:00 or metabolic homeostasis with the 24-hour feedingefasting cycle. In the Zeitgeber Time (ZT) 0) and constant room temperature conditions at well-characterized b-cell, insulin exhibits a diurnal rhythm in secretion the animal facility before use. All experimental procedures were and this pattern in insulin release is more pronounced when nutrients approved by the Animal Care Committee of the University of Toronto. are delivered orally rather than intravenously [15]. This implicates temporal incretin secretion as an essential link between nutrient 2.2. Oral glucose tolerance tests ingestion and deposition through the upregulation of glucose- Oral glucose tolerance tests (OGTTs) were conducted on 4-hour fasted stimulated insulin secretion. mice with their basal blood glucose obtained prior to the administration It was previously reported that the enteroendocrine incretin hormone of an oral gavage of glucose at a concentration of 5 g/kg of body weight glucagon-like peptide-1 (GLP-1) is secreted in a circadian manner [25]. OGTTs were conducted on the C57Bl/6J mice at ZT2, 6, 10, 14, from enteroendocrine L-cells in rats and humans [16e20]. Although 18, and 22. The nighttime studies were carried out under a red light; circadian GLP-1 secretion has been extensively tested in rat models two tests were conducted on most animals with a one-week recovery using physiological disruptors such as constant light and obesogenic interval. Bmal1 WT and KO mice were tested at the trough (ZT2) and feeding [16,18], the lack of appropriate genetically modified animals peak (ZT14) time points of GLP-1 secretion established in the C57Bl/6J has precluded determination of the role of the molecular clock in mice. Blood was collected from the tail vein at 0 min and then 10 and diurnal GLP-1 secretion. Circadian activity has also been shown in the 60 min after the oral gavage to measure the glucose using a OneTouch murine (m) GLUTag and human (h) NCI-H716 L-cell lines, which exhibit meter (LifeScan, Burnaby, BC, Canada) and the plasma GLP-1 and cell-autonomous rhythmic patterns in Bmal1, with GLP-1 secretion insulin levels using a MesoScale Discovery (MSD) assay for the total paralleling Bmal1 expression [16,17,21]. Furthermore, suppression of GLP-1. Bmal1 with palmitate in mGLUTag L-cells is associated with dampened GLP-1 release, while primary intestinal cultures generated from Bmal1 2.3. Cell culture KO mice also demonstrate decreased GLP-1 secretion [18,21]. Male mGLUTag cells were used as a model of intestinal L-cells due to Nonetheless, the molecular mechanism linking Bmal1 expression to their close representation of in vivo GLP-1 secretion [16,21]. The cells circadian GLP-1 secretion remains largely unknown. were grown in DMEM with 25 mmol/L glucose and 10% FBS [16,21]. Interestingly, impaired GLP-1 secretion has been observed in both cell Male hNCI-H716 cells were used as a human L-cell model because and animal models of SNARE deficiency. The SNARE proteins mediate they respond to known GLP-1 secretagogues [17,35]. They were fusion of the secretory granule to the cell membrane, enabling grown in suspension in cell culture flasks with RPMI 1640 medium exocytosis of the granule contents [22,23] and, indeed, the SNARE containing 10% FBS and 100 U/ml penicillin/streptomycin. For the proteins, VAMP2, SYNTAXIN1A, and SYNAPTOTAGMIN-7, have been experiments, hNCI-H716 cells were plated onto cell culture plates demonstrated to play essential roles in GLP-1 secretion [24e26]; coated with 0.5 g/ml Corning Matrigel (Thermo Fisher Scientific, however, it is uncertain if these proteins regulate secretion in a tem- Waltham, MA, USA). poral manner. Evidence from a- and b-cells suggests that SNAREs and their accessory regulators exhibit rhythmic expression [27,28]. Sec- 2.4. In vitro synchronization retagogin (SCGN), a SNARE-regulatory protein [29e31], has been For all of the circadian experiments, the cells were synchronized using identified as rhythmic in these cell types and has been shown to be a previously established protocol [16e18,21]. In brief, the cells were essential for insulin secretion from b-cells [27,28,30,32,33]. SCGN is a grown for two days from the last split and starved in an appropriate calcium-binding protein that interacts with the core SNARE protein media containing 0.5% FBS for 12 h to induce quiescence. The cells SNAP25 and b-actin in b-cells, both of which are also known to be were then synchronized with 20 mM forskolin (SigmaeAldrich, Oak- involved in GLP-1 secretion by L-cells [24,30,32,34]. Given these ville, ON, Canada) for 1 h, after which the media was changed to similarities between b-cells and L-cells, SCGN was identified as a growing media for up to 48 h. While previous reports suggest that potential target linking circadian Bmal1 expression to GLP-1 secretion. certain synchronizing agents generate more robust rhythms than Herein, for the first time, we define a circadian rhythm in GLP-1 others [36], forskolin was used as a synchronizer because it has been secretion in mice, which is dependent on the core clock gene previously shown to elicit a strong circadian response in immortalized

MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). 125 www.molecularmetabolism.com Original Article

L-cells [16e18,21]. Another established synchronizer, 30% FBS, was difference (DCT). The mean, standard error, and statistical analyses also tested with the mGLUTag L-cells, but resulted in significant cell were performed on the DCT data and only converted to relative death after 24 h (unpublished data). expression levels (2^DCT) for presentation in the figures. The total RNA collected from cell lines using a Paris Kit (Thermo Fisher) 2.5. Human and murine primary cell RNA sequencing was reverse-transcribed using 5X All-In-One Reverse Transcriptase Fluorescent-assisted cell sorting (FACS) of L-cells, either via the MasterMix (Applied Biological Materials, Richmond, BC, Canada), and expression of the fluorescent reporter Venus in transgenic mice or quantitative RT-PCR was conducted using a TaqMan Fast Mix Gene staining with fluorescent antibodies in human tissue isolates, was Expression Assay with primers (Thermo Fisher Scientific) as listed in described previously [37]. RNA isolation and sequencing is described Table S1. Gene expression was calculated using the DDCt method in [37,38] and the data were deposited in the NCBI GEO repository [42]. H3f3a (mGLUTag) and Hist1h3a (hNCI-H716) were used as (human, GSE114853; mouse, GSE114913). The studies were con- control genes as they have been previously established to lack ducted in accordance with the principles of the Declaration of Helsinki circadian rhythms [16,17,21]. and good clinical practice. Human ethical approval was provided by Cambridge Central and South Research ethics committees (Ref: 09/ 2.9. Mass spectrometry H0308/24, 16/EE/0338, and 15/EE/0152), the Inserm ethics com- The mGLUTag cells were grown and synchronized as previously mittee, and Agence de la Biomédecine (Ref: PFS16-004). The animal described, and the total protein was extracted at 8 and 20 h after cell research was regulated under the Animals (Scientific Procedures) Act synchronization using TRIzol reagent. The protein extract was analyzed 1986 Amendment Regulations 2012 and conducted following an via mass spectrometry at the SPARC BioCentre, Hospital for Sick ethical review by the University of Cambridge Animal Welfare and Children (Toronto, ON, Canada). The data were analyzed using Scaffold Ethical Review Body. proteome software and the DAVID functional annotation bioinformatics tool. 2.6. hNCI-H716 cell RNA sequencing Total RNA was extracted using an RNeasy Plus Mini Kit (Qiagen) ac- 2.10. Immunoblotting analyses cording to the manufacturer’s instructions and the samples were Protein was collected using a Paris Kit, quantified by a Bradford assay, tested using an Agilent BioAnalyzer to assess their integrity (visual run on 10% polyacrylamide gel, and transferred onto polyvinylidene electropherogram inspection and RIN score). Sequencing libraries difluoride membranes. The membranes were blocked for 1 h in 5% were prepared using an Illumina TruSeq RNA Library Prep Kit v2 with skim milk in Tris-buffered saline with 0.1% Tween (TBS-T, Sigmae polyA selection and run on an Illumina HiSeq 2500 using Rapid Flow Aldrich). The blots were incubated overnight in skim milk TBS-T with Cell v2 at the Harvard University’s Bauer Sequencing Core, with 76 antibodies as listed in Table S2 and washed in TBS-T. Following in- cycle paired end reads as part of a larger multiplex group of barcoded cubation with anti-rabbit IgG secondary antibodies (Cell Signaling, RNA sample libraries. FASTQC v0.11.8 detected no unexpected con- Danvers, MA, USA; Table S2) in skim milk TBS-T, the membranes were ditions in the resulting reads. The reads (minus the last base) were imaged with SignalFire Elite Enhanced chemiluminescent reagent (Cell aligned to human genome assembly 38 and gene models from GEN- Signaling) and visualized on a Kodak imaging system (Eastman Kodak CODE v28 primary assembly using STAR 2.6.0 [39]. Gene read Company, Rochester, NY, USA). When the membranes were re-probed, quantification was performed using RSEM v1.3.1 [40]. Genes with very the blots were first stripped using Restore PLUS Western blotting low total expression (total expected counts < 1, low þ high conditions) stripping buffer (Thermo Fisher Scientific) and washed in TBS-T. were removed and the remainder normalized using the trimmed mean of M-values method [41] as implemented in the edgeR R package. The 2.11. Immunofluorescence log2 counts per million (CPM) were estimated with a prior default of Formalin-fixed, paraffin-embedded murine (UHN Pathology Services, 0.25 CPM per gene. The data were deposited in the NCBI GEO re- Toronto, ON, Canada) and human (OriGene) ileal sections were dew- pository (human, GSE136369). axed, rehydrated, and blocked in 10% normal goat serum (NGS)/PBS for 1 h and then incubated in 10% NGS/PBS with rabbit anti-Scgn (Cell 2.7. Microarray analysis Signaling) and mouse anti-GLP-1 (Abcam, Inc., Toronto, ON, Canada) The RNA was extracted, reverse-transcribed, and subjected to antibodies for 1 h (Table S2), followed by incubation with Alexa Fluor microarray analysis at the Ontario Cancer Institute Genomics Centre 488- and 555-labeled secondary antibodies (Table S2) for 1 h. The (Toronto, ON, Canada) using a mouse WG-6 V2 Illumina BeadChip as immunostained cells were counted and compared based on the previously reported [16]. Gene ontology enrichment data were obtained presence and/or absence of GLP-1 and SCGN co-staining, with the from the Walter and Eliza Hall Institute of Medical Research bioinfor- average percent distribution determined as the number of cells in each matics resource on August 20, 2019 (http://bioinf.wehi.edu.au/ category divided by the total number of cells observed. software/MSigDB/mouse_c5_v5p2.rdata). The data were deposited Cells for immunocytochemistry were grown on Falcon multi-chamber in the NCBI GEO repository (human, GSE136573). microscope slides. For translocation experiments, the cells were synchronized as previously described and then treated with 10 7 M 2.8. Gene expression analyses glucose-dependent insulinotropic polypeptide (GIP, an established Total RNA from FACS-sorted cells [37] was isolated using a Microscale rodent L-cell secretagogue [16e18,21]). Live cells were then incu- RNA Isolation Kit (Ambion) and reverse transcribed according to bated in 2.5 mg/ml of wheat germ agglutinin-Alexa Fluor 488 Conjugate standard protocols. Quantitative RT-PCR was performed with a 7900 (Thermo Fisher) in HBSS at 37 C for 10 min. All of the cells were then HT Fast Real-Time PCR system (Applied Biosystems). The PCR reaction fixed in 4% paraformaldehyde for 30 min at 37 C, permeabilized with mix consisted of first-strand cDNA template, appropriate TaqMan 0.1% Triton X-100 (SigmaeAldrich) in PBS for 20 min, and incubated probe/primer mix, and PCR Master Mix (Thermo Fisher Scientific). The in 1% BSA for 30 min at 37 C, followed by incubation with rabbit anti- expression of Scgn was compared with that of Actb measured on the Scgn antibody (Table S2). All of the cells were then incubated with same sample in parallel on the same plate, demonstrating a CT Alexa Fluor 488-labeled secondary antibody for 1 h (Table S2).

126 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com Pearson’s correlation coefficient (PCC) was used to measure the secretion determined as the 100x the media content of the GLP-1 colocalization of wheat germ agglutinin membrane staining and SCGN divided by the total (media þ cells) GLP-1 content. using the mean intensities of the green and red channels, respectively [43]; a value of zero represents probes that were not correlated with 2.16. Statistical analyses one another. Regions of interest were generated in an unbiased All data are expressed as mean SEM. The delta area under the curve manner by outlining the cell membrane. PCC was calculated using NIS- (DAUC) was calculated using the trapezoidal rule for the changes in Elements imaging software (Nikon Corporation). hormonal and glycemic responses from fasting over the entire 60-min The stained sections and cells were mounted in Vectashield mounting study. In vivo circadian data were tested for significant 24-hour medium containing DAPI (Vector Laboratories, Burlington, ON, Can- rhythms using CircWave software (www.hutlab.nl). In vitro rhythmic ada). Imaging was conducted using a Nikon Swept Field confocal analyses enabling the identification of the significant period was microscope and immunofluorescence analysis was performed with calculated using MetaCycle [46,47] and curves were plotted using a NIS-Elements Imaging. damped sine feature with GraphPad Prism. All other figures were analyzed for significance via ANOVA (1-, 2-, or 3-way) followed by 2.12. Chromatin immunoprecipitation assay Tukey’s test as appropriate. Differential expression analysis of the Non-canonical E-boxes for BMAL1 [44] (CATG(T/C)G) were identified in microarray and protein expression was performed using linear models the 50 Scgn promoter at 672 bp (CATGCG), 1176 bp (CACGCG), and through the R package limma. 1252 bp (CATGTG) upstream of the transcription start site. ChIP was conducted using a SimpleChIP Enzymatic Chromatin IP Kit with 3. RESULTS Magnetic Beads #9003 (Cell Signaling) per the manufacturer’s instructions. In brief, cell protein was cross-linked to DNA with 37% 3.1. Circadian GLP-1 secretion is dependent on the core clock formaldehyde for 10 min at room temperature, digested with micro- gene Bmal1 coccal nuclease for 20 min at 37 C, and incubated with BMAL1 To establish whether GLP-1 secretion follows a circadian rhythm in antibody at 4 C overnight (Table S2). The protein was precipitated mice, 4-hour fasted C57Bl6/J mice were administered an identical oral using Protein G magnetic beads and the DNA was eluted from the glucose load at six time points throughout a 24-hour day. Fasting levels beads using a Magnetic Separation Rack (Cell Signaling). The DNA was of GLP-1, insulin, and blood glucose varied slightly by time of day amplified by PCR using SYBR Green [45] with the listed primers (Figures 1A and S1A). Therefore, to directly compare the L-cell (Table S3). secretory response to the same OGTT over the course of the day, the data were examined as the change from the basal levels. As expected, 2.13. Co-immunoprecipitation plasma GLP-1 increased at 10 min post oral gavage at all times of the Cells were washed with HBSS and lysed with Cell Lysis Buffer (Cell day; however, the peak response was observed at ZT14, which aligns Signaling). The protein (200 mg in 200 ml of lysis buffer) was incubated with the onset of the dark, the feeding period in mice (Figure 1A). with 4 ml of anti-SNAP25 or anti-Scgn antibody (Table S2) overnight Representation of the data as the DAUC generated a curve that with rotation, incubated with Protein A Magnetic Beads (New England significantly (p < 0.05) fit to a 24-hour rhythm, with a peak at ZT14 Biosystems) for 2 h with rotation, and washed with lysis buffer. The and a corresponding trough at ZT2 (Figure 1B). The plasma insulin and beads were then placed in 20 ml of 3X Reducing SDS Loading Buffer blood glucose concentrations also increased at all time points after the (Cell Signaling) and heated at 95 C for 5 min. The protein effluent was oral glucose load (Figure S1A). collected using a Magnetic Separation Rack (Cell Signaling) and loaded Although we previously established that ex vivo adult mouse ileal crypt directly onto a 10% polyacrylamide gel as previously described for cultures generated from Bmal1 KO mice have impaired GLP-1 immunoblotting of SCGN or b-actin, respectively (Table S3). secretory capacity [21], time-dependent GLP-1 secretion had not yet been determined in vivo in these animals. We therefore conducted 2.14. siRNA-mediated knockdown OGTTs at the established trough (ZT2) and peak (ZT14) of the GLP-1 In vitro knockdown studies utilized ON-TARGETplus siRNA for Scgn and secretion in 4 h-fasted Bmal1 KO compared to WT (control) mice. ON-TARGETplus Non-targeting Pool scRNA as a control (Dharmacon Consistent with their known impairment in pancreatic b-cell function Inc., Lafayette, CO, USA). The RNA constructs were used in combi- [9], the KO mice demonstrated impaired insulin release at both time nation with Dharmafect3 using a reverse-transfection protocol as points (Figure S1B), with corresponding hyperglycemia (Figure S1C). recommended by the manufacturer (Dharmacon Inc.). To maintain The WT mice displayed greater GLP-1 secretion at ZT14 than ZT2, knockdown in the synchronization experiments, the media for both confirming the pattern observed in the C57Bl6/J mice. In contrast, the starvation (0.5% FBS) and serum shock (10% FBS with forskolin) steps KO mice demonstrated a loss of the normal peak and trough rhythm, and all subsequent incubations included siRNA or scRNA in combi- with markedly impaired GLP-1 release at the peak (ZT14) time point of nation with Dharmafect3 as appropriate. Cells were then extracted for secretion only (p < 0.05; Figure 1CeD). Taken together, these data qRT-PCR or analyzed for GLP-1 secretion. establish a circadian pattern in GLP-1 secretion in the mouse that depends on the expression of the core clock gene Bmal1. 2.15. GLP-1 secretion assay Cells were incubated for 2 h in DMEM with 0.5% FBS containing 3.2. Identification of secretagogin as a potential regulator of 10 7 M GIP or vehicle alone (control). As previously noted, GIP was circadian GLP-1 secretion used as a secretagogue due to its ability to reliably stimulate temporal To identify targets linking the rhythmic expression of Bmal1 to the GLP-1 release by rodent L-cells [16e18,21]. Peptides in the media circadian secretion of GLP-1, a microarray analysis was conducted on and cells were collected by reversed-phase adsorption using Sep-Pak the synchronized mGLUTag cells at the previously reported peak (4 h) cartridges (Waters Associates, Milford, MA, USA), and the GLP-1 levels and trough (16 h) of Arntl (Bmal1) mRNA expression [16,17] (see also were measured using a Total GLP-1 Radioimmunoassay Kit (Millipore, Figure 4A,F). Although no genes were significantly downregulated, 13 Etobicoke, ON, Canada). All of the data are expressed as percent genes were identified that were significantly upregulated (p < 0.05;

Figure 1: Circadian GLP-1 secretion is dependent on the core clock gene Bmal1. (A-B) OGTTs were conducted on 4-hour fasted C57Bl/6J mice at six time points throughout a 24-hour light (open bars)-dark (closed bars) cycle (ZT0 ¼ 06:00), with individual plots for each time point represented in (A) and the 24-hour DAUC proﬁle represented in (B). (C- D) OGTTs were performed on 4-hour fasted Bmal1 WT and KO mice at ZT2 and ZT14 with individual plots for each time point shown in (C) and the DAUC of each timepoint represented in (D). (A and C)t0 indicates the absolute fasting values for GLP-1 in pg/ml. n ¼ 4e8 mice for all time points in all of the experiments. *p < 0.05.

Table S4), including the positive control Arntl (log2 fold- identified as significantly upregulated (p < 0.001, Figure 2C) at the change ¼ 1.103288, p < 0.05). Furthermore, 34 pathways relating to peak of GLP-1 secretion (8 h). Gene set enrichment analysis vesicle transport were significantly different between the two time (Figure 2D) of the mass spectrometry data identified several pathways points (p < 0.05; Table S5), including the GO transport vesicle matching the microarray findings, including the upregulation of pro- (Figure 2A) and GO SNARE complex (Figure S2), which were both teins related to secretory granules (including SGCN), protein transport, significantly upregulated at the 4-hour time point. The GO transport and the actin cytoskeleton (both including b-actin) at the 8-hour time vesicle includes proteins that move cargo between the ER and Golgi or point. to the membranes for secretion, including the SNARE proteins iden- RNA-seq analysis conducted on intestinal cells collected from murine tified in the GO SNARE complex. This group also includes transcripts Gcg-Venus mice revealed that Scgn was expressed at higher levels in for accessory proteins that facilitate vesicle/granule transport, such as L-cells in the ileum and colon than other intestinal epithelial cells Scgn and Stxbp1. A number of genes in this pathway that are essential (p < 0.001 and p < 0.01, respectively; Figure 3A). Scgn expression for secretion from the L-cells were increased with Bmal1 at the 4-hour was also analyzed by RT-qPCR and compared between colonic Lþ/L- time point (Figure 2B) as well as the transcript for Scgn. In addition, cells and mGLUTag cells, confirming that colonic L-cells express more given the importance of the actin cytoskeleton in GLP-1 secretion [34] secretagogin than non-L-cells (p < 0.05, Figure 3B). In addition, hu- and a recently reported interaction between SCGN and b-actin [32], man jejunal L-cells were compared with enteroendocrine cells (EECs) GO-actin cytoskeleton (Figure S3) was identified as a significantly and other intestinal epithelial cells for SCGN expression, with both L- upregulated pathway at the 4-hour time point (p < 0.05), which in- cells and EECs expressing more secretagogin than the negative con- cludes Actb, Stx1b, Vamp2, and Snap25. Collectively, these findings trols (p < 0.01e0.001, Figure 3C). suggest possible roles for vesicle transport and/or SNARE proteins in Murine (Figure 3D) and human (Figure 3E) ileal sections were also the regulation of circadian GLP-1 secretion. immunostained to assess the co-expression of secretagogin with GLP- þ þ To further screen for targets mediating the temporal secretion of GLP- 1. Analysis of the percent distribution of SCGN and GLP-1 cells þ 1, the proteome of the synchronized mGLUTag cells was studied at the (Figure 3FeG) revealed that SCGN was expressed in all of the GLP-1 peak (8 h) and trough (20 h) time points of GLP-1 secretion [16] (see cells. SCGN was also identified in a significant number of GLP-1- cells. also Figure 8E). Mass spectrometry analysis identified 2,050 proteins However, the murine ileal samples had a higher percentage of þ þ þ with a <5% false-positive detection rate, including 1,749 known SCGN :GLP-1 cells than SCGN GLP-1- cells in the crypts proteins, with 333 proteins upregulated at the 8-hour time point and (p < 0.001), whereas this distribution was reversed in human ileal 414 proteins upregulated at the 20-hour time point. SCGN was tissue (p < 0.001).

128 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com Figure 2: Identification of potential targets regulating circadian GLP-1 secretion. (A) Heat maps of GO transport vesicle and (B) selected L-cell secretory genes identified by microarray in synchronized the mGLUTag L-cells showing log2 fold-change between the two time points, 4 and 16 h (n ¼ 3 for each time point). (C) Volcano plot of the mass spectrometry results of the synchronized mGLUTag L-cells at peak (8 h) and trough (20 h) of GLP-1 secretion (n ¼ 3 for each time point). SCGN is indicated by the open circle. (D) Pathway enrichment analysis of protein clusters that were up regulated at each time point, with SCGN identified under the secretory granules cluster.

To determine the expression of several genes of interest in our cell expression of Per2 (p ¼ 0.06; Figure 4B) was used as a positive con- models, microarray and RNA-seq analyses were conducted on the trol for the synchronization as previously reported [16,21]. Scgn mRNA mGLUTag (Figure 3H) and hNCI-H716 (Figure 3I) cells, respectively. expression peaked at 0 h, with a period of 24 h (p < 0.001; Figure 4C), Transcripts for Scgn/SCGN were expressed in both cell lines and in consistent with a circadian rhythm. To further analyze the rhythmic those for proglucagon (Gcg; the prohormone for GLP-1) and a variety properties of secretagogin, protein was also collected and analyzed for of different known clock (Arntl, Per2, Nr1d1, and Rora) and SNARE the circadian expression of BMAL1 (Figure 4D) and SCGN (Figure 4E). (Stx1a, Vamp2, Snap25, Scgn, and Stxbp1) proteins. Immunostaining Although the pattern in BMAL1 did not reach significance, SCGN further demonstrated that SCGN was localized to both the nucleus demonstrated a significant circadian rhythm (p < 0.001), peaking at 4 h and cytoplasm of the mGLUTag (Figure 3J) and hNCI-H716 with a period of 28 h. Similar analyses of BMAL1 (Figure 4F) and PER2 (Figure 3K) cells. Collectively, these findings demonstrate the (Figure 4G) transcripts over 36 h in the hNCI-H716 cells revealed pre- expression of secretagogin in the murine and human L-cell both viously reported patterns of expression, similarly validating the syn- in vivo and in vitro. chronization of the cells in this model [17]. SCGN also demonstrated strong circadian expression in these cells, peaking at 22 h with a period 3.3. Scgn is expressed in a circadian manner in murine and human of 24 h (p < 0.001, Figure 4H). L-cells To further investigate the rhythmic expression of secretagogin, mRNA 3.4. Temporal interactions of BMAL1 with the Scgn promoter was extracted from synchronized mGLUTag cells every 4 h for 48 h. The To test whether Scgn expression may be driven by BMAL1 binding to circadian expression of Arntl (p < 0.001; Figure 4A) and anti-phasic the Scgn promoter, ChIP analysis was conducted on synchronized

Figure 3: Scgn is expressed in primary and immortalized murine and human L-cells. (A) RNA-seq analysis of Scgn in murine Gcg-Venus L-cells and Venus-negative cells from duodenal, ileal, and colonic sections (n ¼ 2e3 for each cell population). (B) RT-qPCR for Scgn in murine colonic Gcg-Venus L-cells, Venus-negative cells, and mGLUTag L- cells (n ¼ 3 for each cell population). (C) RNA-seq analysis for SCGN in human jejunal L-cells, enteroendocrine cells, and non-endocrine intestinal epithelial cells (n ¼ 11 for each cell population). (D-G) Immunostaining of murine (D) and human (E) ileal sections for SCGN and GLP-1 (representative images of SCGNþ:GLP-1þ (top) and SCGNþ:GLP-1- (bottom) cells are shown). Cell count histogram for murine (F) and human (G) ileal SCGN and/or GLP-1 stained cells (n ¼ 5 sections for each species; a total of w100 cells were counted per section). (H-I) Microarray analysis of mGLUTag cells (n ¼ 6) (H) and RNA-seq analysis of hNCI-H716 cells (n ¼ 2) (I) for proglucagon, clock, and SNARE protein transcripts. (J-K) Immunostaining of mGLUTag (J) and hNCI-H716 (K) L-cells for SCGN; DAPI shows the nuclear stain (representative images of n ¼ 4 are shown). *p < 0.05, **p < 0.01, ***p < 0.001. mGLUTag cells at time points both before and during the peak and 3.5. Secretagogin is recruited to the plasma membrane and binds trough of BMAL1 expression and GLP-1 secretion (at 4 and 8 h vs 16 b-actin but not SNAP25 in a temporal manner and 20 h). Noncanonical E-boxes were identiﬁed in the Scgn pro- To determine whether secretagogin is recruited to the membrane moter at 672, 1176, and 1252 bp upstream of the transcription start following the stimulation of GLP-1 secretion, the synchronized site (Figure 5A). ChIP analysis (Figure 5B) revealed increased binding mGLUTag cells were analyzed by immunocytochemistry to examine at two sites (672 and 1252 bp) during the peak (4e8h) the SGCN localization. At the 8-hour time point, increased localization compared to the trough (16e20 h) time points (p < 0.05), consistent of SCGN at the plasma membrane was observed 60 min after with a role for BMAL1 in the circadian pattern of Scgn expression. stimulation with the known GLP-1 secretagogue, GIP [16,17]

130 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com Figure 4: Scgn expression is circadian in mGLUTag and hNCI-H716 L-cells. (A-C) mRNA and (D-E) protein expression over 48 h in synchronized mGLUTag L-cells for Bmal1 (A), Per2 (B), Scgn (C), BMAL1 (D), and SCGN (E;n¼ 8, conducted as 4 replicates from each of 2 independent splits; representative blots are shown in (D-E)). (F-G) mRNA expression over 36 h in synchronized hNCI-H716 cells for BMAL1 (F), PER2 (G), and SCGN (H;n¼ 8, conducted as 4 replicates from each of 2 independent splits).

(Figure 6AeB). In contrast, at the 20-hour time point, decreased recruitment was observed 60 min after stimulation (Figure 6CeD). To establish whether SCGN interacts with the SNARE machinery and/or b-actin in L-cells as previously reported in b-cells [30,32,48], the synchronized mGLUTag cells were treated for 2 h with GIP at both the 8- and 20-hour time points. SNAP25 was then immunoprecipitated (IP) and the blots were probed for co-IP of SCGN (Figure 7A); alternatively, SCGN was pulled-down and the blots were probed for b-actin (Figure 7B). Although SCGN was found to co-IP with SNAP25, no differences in interaction were demonstrated based on the clock time or with GIP stimulation. However, not only was SCGN found to also interact with b-actin, but the amount of b-actin bound to SCGN increased with stimulation at the 8-hour time point only (p < 0.05). Together, these ﬁndings are consistent with b-actin’s role in SCGN translocation to the plasma membrane during GLP-1 secretion.

3.6. Scgn is essential for peak GLP-1 secretion To investigate the functional importance of secretagogin in GLP-1 Figure 5: Time-dependent binding of BMAL1 to the Scgn promoter in mGLUTag secretion, a GLP-1 secretion experiment was conducted in the syn- L-cells. (A) Three noncanonical BMAL1 E-boxes were identiﬁed at 1252, 1176, and chronized mGLUTag cells following knockdown of Scgn. Sc- and siRNA 672 base pairs upstream of the transcription start site. (B) ChIP analysis for the BMAL1 binding sites in the Scgn promoter at 4, 8, 16, and 20 h in synchronized mGLUTag L- treatments had no effect on synchronization, as shown by the expected cells (n ¼ 6, conducted as 3 replicates from each of 2 independent splits). *p < 0.05; anti-phasic expression of Bmal1 and Per2 at the peak and trough time ###p < 0.001 vs negative control (-IgG). points (Figure 8AeB). However, secretagogin knockdown was signiﬁcant

Figure 6: Time-dependent recruitment of SCGN to the cell membrane in mGLUTag L-cells. (A-B) 8 h and (C-D) 20 h after synchronization, mGLUTag L-cells were treated with 107 M GIP or vehicle for 0, 10, and 60 min, followed by staining for SCGN and a membrane marker (with wheat germ agglutinin). Co-localization of SCGN with the cell membrane was determined by Pearson’s correlation coefﬁcient. (n ¼ 4, conducted as 2 replicates from each of 2 independent splits; representative images are shown). *p < 0.05. at both the mRNA and protein levels at 8 and 20 h (p < 0.05, Figure 8Ce secretion in response to GIP at the 8-hour time point (p < 0.01) but D), while the expression of transcripts for other key SNARE proteins was had no effect at 20 h, demonstrating secretagogin’s role in the circadian unaffected (Figure S4). A GLP-1 secretion assay was then conducted at secretion of GLP-1. This loss of response to GIP was observed although the peak (8 h) and trough (20 h) time points under vehicle and GIP- Gipr mRNA expression was actually elevated at 20 h compared to 8 h stimulated conditions. Synchronization of the cells was further after cell synchronization (Figure S5). conﬁrmed by the demonstration of higher GLP-1 secretion in response to GIP at the 8-hour time point compared to the 20-hour time point as 4. DISCUSSION previously reported [16,21] in the scRNA-treated cells (p < 0.05, Figure 8E). Scgn knockdown had no effect on the basal GLP-1 secretion While circadian rhythms are primarily caused by light, peripheral at either time point. However, Scgn knockdown decreased GLP-1 metabolic tissues can be entrained by food intake. As shift workers

Figure 7: Time-dependent interactions of SCGN with b-actin but not SNAP25 in mGLUTag L-cells. mGLUTag L-cells were synchronized and then 8 or 20 h later treated with 107 M GIP or vehicle for 2 h. (A) Immunoprecipitation of SNAP25 and immunoblotting for SCGN. (B) Immunoprecipitation of SCGN and immunoblotting for b-actin. (n ¼ 4, conducted as 2 replicates from each of 2 independent splits; representative blots are shown). *p < 0.05.

132 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com Figure 8: Scgn is essential for peak GLP-1 secretion in mGLUTag L -cells. mGLUTag L-cells were treated with scRNA or Scgn siRNA, synchronized and then 8 or 20 h later treated with 107 M GIP or vehicle for 2 h. The cells were then analyzed for Bmal1 (A), Per2 (B), Scgn (C), SCGN (a representative blot is shown) (D), and GLP-1 secretion (E) (n ¼ 8, conducted as 4 replicates from each of 2 independent splits). *p < 0.05, **p < 0.01, ***p < 0.001.

have a higher incidence of obesity and type 2 diabetes [49e51], these Previous studies of rats identified a circadian rhythm in the GLP-1 epidemiological data implicate diurnal insulin patterns in disease. The secretory response to an oral glucose load that peaked at ZT10, incretin hormones account for approximately 50% of insulin secretion just prior to the onset of their dark or active/feeding period [16]. The after a meal [52,53], and secretion by the b-cells is coordinated, at mouse model also exhibited circadian GLP-1 secretion in response least in part, by circadian rhythms in GLP-1 release [16]. However, to the same stimulus; however, the peak of secretion was slightly although GLP-1 secretion by mGLUTag L-cells has been shown to shifted, occurring at ZT14 as the mice entered their feeding period. parallel Bmal1 expression [16], and primary intestinal cultures from These findings in rodents are consistent in that peak GLP-1 secre- Bmal1 KO mice show decreased GLP-1 secretion ex vivo [21], the tion in both species occurs as an anticipatory response to increased effect of Bmal1 KO on diurnal GLP-1 secretion in vivo remained un- food intake occurring throughout the dark period. Interestingly, known. Furthermore, the exact mechanism by which Bmal1 regulates although obesogenic feeding in rats disrupts the rhythm in GLP-1 circadian GLP-1 release is also unclear, although our previous release such that the normal trough of secretion at the onset of research demonstrated roles of thyrotrophic embryonic factor and the light period is lost [18], KO of Bmal1 in mice not only disrupted protein tyrosine phosphatase 4a1 in regulating the peak of GLP-1 rhythmic GLP-1 release, but also impaired GLP-1 secretion at the release [16]. We have also shown the importance of the Bmal1- peak time point only. Whether these specific differences in timing nicotinamide phosphoribosyltransferase (NAMPT) pathway, identi- are consequent to the nature of the circadian disruptors utilized and/ fying mitochondrial activity and ATP-dependent cellular metabolism as or represent species-dependent differences remain unknown. essential for peak GLP-1 secretion [21]. The results of the present However, similar differences have been noted in humans, wherein a study demonstrate that the accessory SNARE protein, secretagogin, is phase delay of 3 27 h has been reported to impair GLP-1 release, not only a target of Bmal1, but also plays an essential role in regulating whereas a 9-hour phase advance has been reported to have no the peak, but not the trough, of circadian GLP-1 secretion. effect [54,55].

Previous studies demonstrated key roles for several SNARE proteins in stimulated compared to basal GLP-1 secretion [25,34]. Similar studies of GLP-1 secretion (VAMP2, STX1A, and SYT7) [24e26]. Given the NIT-1 b-cells also showed that only second-phase insulin secretion was demonstrated rhythmic expression of SNARE proteins in other endocrine altered following Scgn silencing [32]. Indeed, cell stimulation is known to cell types, such as islet a-andb-cells [28], they are ideally situated to be essential for SGCN’s role as a calcium sensor, inducing conforma- provide a mechanistic link between circadian Bmal1 expression and tional changes that are required for the facilitation of secretion [29e GLP-1 secretion. Furthermore, isolated L-cells from mouse models of 31,48,62,63]. In the b-cells, this appears to be due, at least in part, elevated GLP-1 secretion demonstrate changes in vesicle organization to inhibition by tomosyn, which dissociates from SCGN when exposed to and vesicle localization [56]. Consistent with this evidence, our tran- calcium, liberating SCGN to interact with its secretory partners, including scriptomic and proteomic findings show that the pathways related to actin and SNAP25 [30]. Consistent with this, the secretagogue used in both vesicle transport and the SNARE proteins are upregulated at the the present study, GIP, increases the cAMP levels in the L-cells, which is þ peak GLP-1 secretion time point. SCGN was also found to co-IP with b- known to enhance intracellular Ca2 concentrations and lead to actin as well as with the SNARE protein SNAP25, consistent with find- increased GLP-1 secretion [64]. Finally, other SNARE regulators that ings in other cell types, including b-cells [30,48,57]. have been suggested to be rhythmic [28] and are known to be essential Actin cytoskeletal pathways were found to be upregulated in the for insulin secretion [65,66]werealsoidentified in the mGLUTag L-cell mGLUTag L-cells at the peak secretory time point by both microarray proteome as being upregulated at the peak secretion time point, and mass spectrometry analysis. This is consistent with studies showing including the SNARE-accessory protein STXBP-1 (Munc-18). Further that the actin cytoskeleton plays an essential role in hormone secretion investigation into the role of other potential mediators of circadian GLP-1 by the L-cells and b-cells, wherein actin remodeling upon stimulation is secretion is thus warranted. necessary to permit the stimulation of granule exocytosis [34,58e61]. It Importantly, SCGN was identified in all primary murine and human L- is therefore possible that secretagogin plays a role in this remodeling to cells. Given recent evidence demonstrating a diversity of L-cells along regulate GLP-1 secretion in a temporal manner. In line with this evi- the crypt villus axis [67], L-cell SCGN expression appears to be dence, SCGN has been shown to be important in the organization of the ubiquitous, suggesting that it plays an essential role in L-cell function. actin cytoskeleton, interacting with trafficking proteins and regulating Secretagogin may also influence the development of L-cells, as evi- focal adhesion [32,62]. In the b-cells, this interaction of SCGN with actin dence shows that it plays an essential role in b-cell development and has been shown to increase with stimulation [62], a finding that was causes a-cell hyperplasia when knocked out [33]. Recent reports also reproduced in our studies in the L-cells, with stimulation causing implicate reduced secretagogin expression in type 2 diabetes [68e increased secretagogin binding to b-actin at the 8-hour time point but 70], consistent with a report that whole body SCGN KO mice are not at 20 h. Further investigation is required to determine if the temporal glucose intolerant [69]. Given the importance of GLP-1 and, subse- b-actin-secretagogin interaction directly affects GLP-1 secretion. How- quently, insulin secretion for the maintenance of glucose homeostasis, ever, interestingly, as the interaction with b-actin was found to be time- it is possible that some of the effects of altered secretagogin dependent, whereas that with SNAP25 was not, these findings suggest expression may be L-cell mediated. an active role for SGCN in the translocation of GLP-1-filled secretory It is acknowledged that the Scgn knockdown studies presented herein granules to the cell membrane and a more permissive role in SNARE- were only conducted in vitro. Although the cellular models utilized are mediated exocytosis. Further evidence for this is provided by the known to be representative of primary L-cells in terms of their GLP-1 demonstration that only peak GLP-1 secretion by the mGLUTag L-cells secretion and response to secretagogues [16,21,35,71], one draw- was associated with increased translocation of SGCN to the cell back to in vitro circadian experiments is that in vivo rhythms are membrane. This is also in agreement with the decreased recruitment of orchestrated by a range of cues, with L-cells synchronized by nutrient Scgn to the cell membrane observed at the 20-hour time point, which intake and a variety of hormones, neural inputs, cytokines, and the potentially explains the trough GLP-1 secretory response observed in this microbiome. Thus, in vitro experiments are subject to a slow loss of study and in other publications [16e18,21]. synchronicity given the absence of such cues, limiting experiments to Expressed in a rhythmic manner in both the mGLUTag and hNCI-H716 48-hour time intervals. Nonetheless, while our cells may lack these cells, Scgn transcript levels paralleled those of SCGN with a 4-hour cues, synchronization protocols have been established to mimic these translational delay. The Scgn expression patterns were also more environments, driving these rhythms as seen in other tissues and cell consistent with those of Bmal1 in both the murine and human cells, types [16e18,21,27,28,72,73]. Additionally, our studies used only rather than with that of Per2, which was anti-phasic to Bmal1 and Scgn single synchronization protocol (i.e., with forskolin), although studies of in the mGLUTag L-cells, but arrhythmic in the hNCI-H716 L-cells (pre- b-cells have shown that multiple different synchronizing agents that sent study and [16,17]). These findings further confirmed Bmal1 as a also act through the cAMP-PKA pathway generate similar effects on the prime target regulating circadian GLP-1 secretion, as was also circadian expression of a Per2-luciferase construct [36]. Finally, GIP was demonstrated by ChIP analysis, demonstrating that the Scgn promoter is used as the sole secretagogue in the present study, suggesting that the a direct target of BMAL1, with greater binding of BMAL1 to 2 E-boxes in results may apply solely to GIP-stimulated GLP-1 secretion; however, Scgn at the peak compared to the trough time point of GLP-1 secretion. previous studies directly compared the temporal responses to GIP with Interestingly, in vivo loss of Bmal1 and in vitro knockdown of Scgn those of several other L-cell secretagogues (insulin and bethanechol) in resulted in a similar phenotype of loss in GLP-1 secretory rhythm and the mGLUTag L-cells, finding no differences between the 3 agents with impaired secretion only at the peak time point, further suggesting a tight respect to the peak and trough GLP-1 secretory responses [16e18,21]. interplay between Bmal1 and Scgn. Furthermore, consistent with an In conclusion, we have established the circadian pattern of GLP-1 essential role of SGCN in circadian GLP-1 secretion, only peak GLP-1 secretion in mice and shown that it is dependent on the core clock release was reduced by Scgn knockdown. Interestingly, this effect gene Bmal1. We have also characterized secretagogin expression in was observed only with respect to the stimulated GLP-1 release at this murine and human L-cells for the first time, identifying secretagogin as time point, with no observed change in the basal secretion. These a novel regulator of circadian GLP-1 secretion: under the control of findings are consistent with studies showing that disruption of the actin rhythmic BMAL1 expression, SCGN binds to b-actin, is recruited to the cytoskeleton and syntaxin-1A knockout cause impairments solely in cell membrane in a time-dependent manner that is parallel to its

134 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com circadian expression and is required for stimulated-GLP-1 secretion. [2] Gerhart-Hines, Z., Lazar, M.A., 2015. Circadian metabolism in the light of Due to the insulinotropic effects of GLP-1, both long-acting GLP-1 evolution. Endocrine Reviews 36(3):289e304. https://doi.org/10.1210/ receptor agonists and GLP-1 degradation inhibitors are used as a er.2015-1007. therapy for type 2 diabetes [74]. SNARE regulators may provide a novel [3] Mohawk, J.A., Green, C.B., Takahashi, J.S., 2012. Central and peripheral target for type 2 diabetes therapies given their regulation of the circadian clocks in mammals. Annual Review of Neuroscience 35(1):445e secretion of key metabolic hormones, including not only GLP-1, but 462. https://doi.org/10.1146/annurev-neuro-060909-153128. also insulin. Further investigation of the developing link between [4] Zhang, R., Lahens, N.F., Ballance, H.I., Hughes, M.E., Hogenesch, J.B., 2014. circadian rhythms, diabetes, and obesity [49e51] and improved un- A circadian gene expression atlas in mammals: implications for biology and derstanding of the molecular drivers behind the circadian secretion of medicine. Proceedings of the National Academy of Sciences of the United States GLP-1 may therefore have therapeutic implications, including time- of America 111(45):16219e16224. https://doi.org/10.1073/pnas.1408886111. sensitive therapies. [5] Tahara, Y., Shibata, S., 2016. Circadian rhythms of liver physiology and Dis- ease: experimental and clinical evidence. Nature Reviews Gastroenterology CONTRIBUTIONS and Hepatology 13(4):217e226. https://doi.org/10.1038/nrgastro.2016.8. [6] Johnston, J.D., Ordovás, J.M., Scheer, F.A., Turek, F.W., 2016. Circadian ADB and AM designed and conducted the studies, performed the ana- rhythms, metabolism, and chrononutrition in rodents and humans. Advances in lyses, and wrote the paper. EM, PG, EM, JC, AD, JCM, AEA, FR, FMG, Nutrition 7(2):399e406. https://doi.org/10.3945/an.115.010777. and MG-L conducted the studies and performed the analyses. BJC [7] Hoogerwerf, W.A., Hellmich, H.L., Cornélissen, G., Halberg, F., Shahinian, V.B., conducted the analyses. PLB designed the studies, conducted the an- Bostwick, J., et al., 2007. Clock gene expression in the murine gastrointestinal alyses, and wrote the paper. All authors approved the final manuscript. tract: endogenous rhythmicity and effects of a feeding regimen. Gastroen- terology 133(4):1250e1260. https://doi.org/10.1053/j.gastro.2007.07.009. DISCLOSURES [8] Mayeuf-Louchart, A., Staels, B., Duez, H., 2015. Skeletal muscle functions around the clock. Diabetes, Obesity and Metabolism 17:39e46. https:// ADB, AM, EM, PG, EM, JC, AD, JCM, AEA, FR, FMG, MG-L, and PLB doi.org/10.1111/dom.12517. have nothing to disclose. [9] Marcheva, B., Ramsey, K.M., Buhr, E.D., Kobayashi, Y., Su, H., Ko, C.H., et al., ACKNOWLEDGMENTS 2010. Disruption of the clock components CLOCK and BMAL1 leads to hypoinsulinaemia and diabetes. Nature 466(7306):627e631. https://doi.org/ The authors are grateful to Caterina Pipino (Joslin Diabetes Center, Boston, MA, USA) 10.1038/nature09253. for preparing the hNCI-H716 L-cells for sequencing, and to Pierre Larraufie, Paul [10] Rey, G., Cesbron, F., Rougemont, J., Reinke, H., Brunner, M., Naef, F., 2011. fi Richards, Raphaël Scharfmann, and Geoffrey Roberts (University of Cambridge, Genome-wide and phase-speci c DNA-binding rhythms of BMAL1 control Cambridge, UK) for their contributions in generating the mouse and human RNA-seq circadian output functions in mouse liver. PLoS Biology 9(2):e1000595. https:// data. ADB was supported by a graduate studentship from the Banting and Best doi.org/10.1371/journal.pbio.1000595. Diabetes Centre, University of Toronto, Toronto, ON, Canada. AM was supported by a [11] Koike, N., Yoo, S.-H., Huang, H.-C., Kumar, V., Lee, C., Kim, T.-K., et al., 2012. Canada Graduate Doctoral Scholarship from the Canadian Institutes of Health Transcriptional architecture and Chromatin landscape of the core circadian e Research and by graduate awards from the Ontario Graduate Scholarship and clock in mammals. Science 338(6105):349 354. https://doi.org/10.1126/ Banting and Best Diabetes Centre, University of Toronto. EM was supported by a science.1226339. summer studentship from the Banting and Best Diabetes Centre (University of Tor- [12] McCarthy, J.J., Andrews, J.L., McDearmon, E.L., Campbell, K.S., Barber, B.K., fi onto). PG was supported by summer studentships from the Banting and Best Dia- Miller, B.H., et al., 2007. Identi cation of the circadian transcriptome in adult e betes Centre (University of Toronto) and the Banting Research Foundation. BC was mouse skeletal muscle. Physiological Genomics 31(1):86 95. https://doi.org/ supported by a Tier II Canada Research Chair. PLB was supported by a Tier I Canada 10.1152/physiolgenomics.00066.2007. Research Chair. Studies in the Brubaker laboratory were supported by an operating [13] Zvonic, S., Ptitsyn, A.A., Conrad, S.A., Scott, L.K., Floyd, Z.E., Kilroy, G., et al., grant from the Canadian Institutes of Health Research (PJT-15308), and some of the 2006. Characterization of peripheral circadian clocks in adipose tissues. e equipment used in this study was provided by the 3D (Diet, Digestive Tract, and Diabetes 55(4):962 970. https://doi.org/10.2337/diabetes.55.04.06.db05- Disease) Centre funded by the Canadian Foundation for Innovation and Ontario 0873. Research Fund (project numbers 19442 and 30961). Work in the Reimann/Gribble [14] Gachon, F., Loizides-Mangold, U., Petrenko, V., Dibner, C., 2017. Glucose laboratories was supported by the Wellcome Trust (106262/Z/14/Z and 106263/Z/ homeostasis: regulation by peripheral circadian clocks in rodents and humans. e 14/Z) and the UK Medical Research Council (MRC_MC_UU_12012/3). Endocrinology 158(5):1074 1084. https://doi.org/10.1210/en.2017-00218. [15] Kalsbeek, A., Strubbe, J.H., 1998. Circadian control of insulin secretion is CONFLICT OF INTEREST independent of the temporal distribution of feeding. Physiology & Behavior 63(4):553e560. https://doi.org/10.1016/S0031-9384(97)00493-9. None declared. [16] Gil-Lozano, M., Mingomataj, E.L., Wu, W.K., Ridout, S.A., Brubaker, P.L., 2014. Circadian secretion of the intestinal hormone GLP-1 by the rodent L cell. APPENDIX A. SUPPLEMENTARY DATA Diabetes 63(11):3674e3685. https://doi.org/10.2337/db13-1501. [17] Gil-Lozano, M., Hunter, P.M., Behan, L.-A., Gladanac, B., Casper, R.F., Supplementary data to this article can be found online at https://doi.org/10.1016/j. Brubaker, P.L., 2015. Short-term sleep deprivation with nocturnal light molmet.2019.11.004. exposure alters time-dependent glucagon-like peptide-1 and insulin secretion in male volunteers. American Journal of Physiology e Endocrinology and REFERENCES Metabolism 310(1):41e50. https://doi.org/10.1152/ajpendo.00298.2015. [18] Gil-Lozano, M., Wu, W.K., Martchenko, A., Brubaker, P.L., 2016. High-fat diet [1] Bass, J., Lazar, M.A., 2016. Circadian time signatures of fitness and Disease. and palmitate alter the rhythmic secretion of glucagon-like peptide-1 by the Science (New York, N.Y.) 354(6315):994e999. https://doi.org/10.1126/ rodent L-cell. Endocrinology 157(2):586e599. https://doi.org/10.1210/ science.aah4965. en.2015-1732.

[19] Mingrone, G., Nolfe, G., Castagneto Gissey, G., Iaconelli, A., Leccesi, L., remodeling and glucagon-like peptide-1 secretion in the intestinal endocrine L Guidone, C., et al., 2009. Circadian rhythms of GIP and GLP1 in glucose- cell. Endocrinology 150(12):5249e5261. https://doi.org/10.1210/en.2009-0508. tolerant and in type 2 diabetic patients after biliopancreatic diversion. Dia- [35] Gagnon, J., Brubaker, P.L., 2015. NCI-H716 cells. The impact of food bio- betologia 52(5):873e881. https://doi.org/10.1007/s00125-009-1288-9. actives on Health: in vitro and ex vivo models. Springer. p. 221e8. [20] Lindgren, O., Mari, A., Deacon, C.F., Carr, R.D., Winzell, M.S., Vikman, J., [36] Petrenko, V., Dibner, C., 2018. Cell-specific resetting of mouse islet cellular et al., 2009. Differential islet and incretin hormone responses in morning clocks by glucagon, glucagon-like peptide 1 and somatostatin. Acta Physi- versus afternoon after standardized meal in healthy men. The Journal of ologica 222(4):e13021. https://doi.org/10.1111/apha.13021. Clinical Endocrinology and Metabolism 94(8):2887e2892. https://doi.org/ [37] Reimann, F., Habib, A.M., Tolhurst, G., Parker, H.E., Rogers, G.J., 10.1210/jc.2009-0366. Gribble, F.M., 2008. Glucose sensing in L cells: a primary cell study. Cell [21] Martchenko, A., Oh, R.H., Wheeler, S.E., Gurges, P., Chalmers, J.A., Metabolism 8(6):532e539. https://doi.org/10.1016/j.cmet.2008.11.002. Brubaker, P.L., 2018. Suppression of circadian secretion of glucagon-like [38] Roberts, G.P., Larraufie, P., Richards, P., Kay, R.G., Galvin, S.G., peptide-1 by the saturated fatty acid, palmitate. Acta Physiologica 222(4): Miedzybrodzka, E.L., et al., 2019. Comparison of human and murine enter- e13007. https://doi.org/10.1111/apha.13007. oendocrine cells by transcriptomic and peptidomic profiling. Diabetes 68(5): [22] Jahn, R., Scheller, R.H., 2006. SNAREs d engines for membrane fusion. 1062e1072. https://doi.org/10.2337/db18-0883. Nature Reviews Molecular Cell Biology 7(9):631e643. https://doi.org/10.1038/ [39] Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., et al., nrm2002. 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics (Oxford, En- [23] Südhof, T.C., 2014. The molecular machinery of neurotransmitter release gland) 29(1):15e21. https://doi.org/10.1093/bioinformatics/bts635. (nobel lecture). Angewandte Chemie International Edition 53(47):12696e [40] Li, B., Dewey, C.N., 2011. RSEM: accurate transcript quantification from RNA- 12717. https://doi.org/10.1002/anie.201406359. seq data with or without a reference genome. BMC Bioinformatics 12(1):323. [24] Li, S.K., Zhu, D., Gaisano, H.Y., Brubaker, P.L., 2014. Role of vesicle- https://doi.org/10.1186/1471-2105-12-323. associated membrane protein 2 in exocytosis of glucagon-like peptide-1 [41] Robinson, M.D., Oshlack, A., 2010. A scaling normalization method for dif- from the murine intestinal L cell. Diabetologia 57(4):809e818. https://doi.org/ ferential expression analysis of RNA-seq data. Genome Biology 11(3):R25. 10.1007/s00125-013-3143-2. https://doi.org/10.1186/gb-2010-11-3-r25. [25] Wheeler, S.E., Stacey, H.M., Nahaei, Y., Hale, S.J., Hardy, A.B., Reimann, F., [42] Livak, K.J., Schmittgen, T.D., 2001. Analysis of relative gene expression data et al., 2017. The SNARE protein syntaxin-1a plays an essential role in biphasic using real-time quantitative PCR and the 2DDCT method. Methods 25(4): exocytosis of the incretin hormone glucagon-like peptide 1. Diabetes 66(9): 402e408. https://doi.org/10.1006/meth.2001.1262. 2327e2338. https://doi.org/10.2337/db16-1403. [43] Dunn, K.W., Kamocka, M.M., McDonald, J.H., 2011. A practical guide to [26] Gustavsson, N., Wang, Y., Kang, Y., Seah, T., Chua, S., Radda, G.K., et al., evaluating colocalization in biological microscopy. American Journal of Phys- 2011. Synaptotagmin-7 as a positive regulator of glucose-induced glucagon- iology e Cell Physiology 300(4):C723eC742. https://doi.org/10.1152/ like peptide-1 secretion in mice. Diabetologia 54(7):1824e1830. https:// ajpcell.00462.2010. doi.org/10.1007/s00125-011-2119-3. [44] Yoshitane, H., Ozaki, H., Terajima, H., Du, N.-H., Suzuki, Y., Fujimori, T., et al., [27] Perelis, M., Marcheva, B., Ramsey, K.M., Schipma, M.J., Hutchison, A.L., 2014. CLOCK-controlled polyphonic regulation of circadian rhythms through Taguchi, A., et al., 2015. Pancreatic b cell enhancers regulate rhythmic canonical and noncanonical E-boxes. Molecular and Cellular Biology 34(10): transcription of genes controlling insulin secretion. Science (New York, N.Y.) 1776e1787. https://doi.org/10.1128/MCB.01465-13. 350(6261):aac4250. https://doi.org/10.1126/science.aac4250. [45] Loganathan, N., Salehi, A., Chalmers, J.A., Belsham, D.D., 2019. Bisphenol A [28] Petrenko, V., Saini, C., Giovannoni, L., Gobet, C., Sage, D., Unser, M., et al., alters Bmal1, Per2, and rev-erba mRNA and requires Bmal1 to increase 2017. Pancreatic a- and b-cellular clocks have distinct molecular properties neuropeptide Y expression in hypothalamic neurons. Endocrinology 160(1): and impact on islet hormone secretion and gene expression. Genes and 181e192. https://doi.org/10.1210/en.2018-00881. Development 31(4):383e398. https://doi.org/10.1101/gad.290379.116. [46] Hughes, M.E., Hogenesch, J.B., Kornacker, K., 2010. JTK_CYCLE: an efficient [29] Wagner, L., Oliyarnyk, O., Gartner, W., Nowotny, P., Groeger, M., Kaserer, K., nonparametric algorithm for detecting rhythmic components in genome-scale et al., 2000. Cloning and expression of secretagogin, a novel neuroendocrine- data sets. Journal of Biological Rhythms 25(5):372e380. https://doi.org/ and pancreatic islet of langerhans-specificCa2þ -binding protein. Journal of 10.1177/0748730410379711. Biological Chemistry 275(32):24740e24751. https://doi.org/10.1074/ [47] Wu, G., Anafi, R.C., Hughes, M.E., Kornacker, K., Hogenesch, J.B., 2016. jbc.M001974200. MetaCycle: an integrated R package to evaluate periodicity in large scale data. [30] Ferdaoussi, M., Fu, J., Dai, X., Manning Fox, J.E., Suzuki, K., Smith, N., et al., Bioinformatics 32(21):3351e3353. https://doi.org/10.1093/bioinformatics/ 2017. SUMOylation and calcium control syntaxin-1A and secretagogin btw405. sequestration by tomosyn to regulate insulin exocytosis in human ß cells. [48] Rogstam, A., Linse, S., Lindqvist, A., James, P., Wagner, L., Berggård, T., Scientific Reports 7(1):248. https://doi.org/10.1038/s41598-017-00344-z. 2007. Binding of calcium ions and SNAP-25 to the hexa EF-hand protein [31] Maj, M., Wagner, L., Tretter, V., 2019. 20 Years of secretagogin: exocytosis secretagogin. Biochemical Journal 401(1):353e363. https://doi.org/10.1042/ and beyond. Frontiers in Molecular Neuroscience 12:29. https://doi.org/ BJ20060918. 10.3389/fnmol.2019.00029. [49] Fonken, L.K., Nelson, R.J., 2014. The effects of light at night on circadian [32] Yang, S.-Y., Lee, J.-J.J.-H., Lee, J.-J.J.-H., Lee, K.-J., Oh, S.H., Lim, Y.-M., clocks and metabolism. Endocrine Reviews 35(4):648e670. https://doi.org/ et al., 2016. Secretagogin affects insulin secretion in pancreatic b-cells by 10.1210/er.2013-1051. regulating actin dynamics and focal adhesion. Biochemical Journal 473(12): [50] Pan, A., Schernhammer, E.S., Sun, Q., Hu, F.B., 2011. Rotating night shift 1791e1803. https://doi.org/10.1042/BCJ20160137. work and risk of type 2 diabetes: two prospective cohort studies in women. [33] Malenczyk, K., Szodorai, E., Schnell, R., Lubec, G., Szabs, G., H4kfelt, T., PLoS Medicine 8(12):e1001141. https://doi.org/10.1371/journal.pmed. et al., 2018. Secretagogin protects Pdx1 from proteasomal degradation to 1001141. control a transcriptional program required for b cell specification. Molecular [51] Froy, O., 2010. Metabolism and circadian rhythmsdimplications for obesity. Metabolism 14:108e120. https://doi.org/10.1016/j.molmet.2018.05.019. Endocrine Reviews 31(1):1e24. https://doi.org/10.1210/er.2009-0014. [34] Lim, G.E., Xu, M., Sun, J., Jin, T., Brubaker, P.L., 2009. The rho guanosine 50- [52] Nauck, M., 2016. Incretin therapies: highlighting common features and dif- triphosphatase, cell division cycle 42, is required for insulin-induced actin ferences in the modes of action of glucagon-like peptide-1 receptor agonists

136 MOLECULAR METABOLISM 31 (2020) 124e137 Ó 2019 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). www.molecularmetabolism.com and dipeptidyl peptidase-4 inhibitors. Diabetes Obesity and Metabolism, 203e towards thermal and urea denaturation. PLoS One 11(11):e0165709. https:// 216. https://doi.org/10.1111/dom.12591. doi.org/10.1371/journal.pone.0165709. [53] Wang, Z., Wang, R.M., Owji, A.A., Smith, D.M., Ghatei, M.A., Bloom, S.R., [64] Simpson, A.K., Ward, P.S., Wong, K.Y., Collord, G.J., Habib, A.M., Reimann, F., 1995. Glucagon-like peptide-1 is a physiological incretin in rat. Journal of et al., 2007. Cyclic AMP triggers glucagon-like peptide-1 secretion from the Clinical Investigation 95(1):417e421. https://doi.org/10.1172/JCI117671. GLUTag enteroendocrine cell line. Diabetologia 50(10):2181e2189. https:// [54] Gonnissen, H.K.J., Rutters, F., Mazuy, C., Martens, E.A.P., Adam, T.C., doi.org/10.1007/s00125-007-0750-9. Westerterp-Plantenga, M.S., 2012. Effect of a phase advance and phase delay [65] Verhage, M., Maia, A.S., Plomp, J.J., Brussaard, A.B., Heeroma, J.H., of the 24-h cycle on energy metabolism, appetite, and related hormones. The Vermeer, H., et al., 2000. Synaptic assembly of the brain in the absence of American Journal of Clinical Nutrition 96(4):689e697. https://doi.org/10.3945/ neurotransmitter secretion. Science 287(5454):864e869. https://doi.org/ ajcn.112.037192. 10.1126/science.287.5454.864. [55] Hampton, S.M., Morgan, L.M., Lawrence, N., Anastasiadou, T., Norris, F., [66] Oh, E., Kalwat, M.A., Kim, M.-J., Verhage, M., Thurmond, D.C., 2012. Deacon, S., et al., 1996. Postprandial hormone and metabolic responses in Munc18-1 regulates first-phase insulin release by promoting granule docking simulated shift work. Journal of Endocrinology 151(2):259e267. https:// to multiple syntaxin isoforms. The Journal of Biological Chemistry 287(31): doi.org/10.1677/joe.0.1510259. 25821e25833. https://doi.org/10.1074/jbc.M112.361501. [56] Arora, T., Akrami, R., Pais, R., Bergqvist, L., Johansson, B.R., Schwartz, T.W., [67] Beumer, J., Artegiani, B., Post, Y., Reimann, F., Gribble, F., Nguyen, T.N., et al., 2018. Microbial regulation of the L cell transcriptome. Scientific Reports et al., 2018. Enteroendocrine cells switch hormone expression along the crypt- 8(1):1207. https://doi.org/10.1038/s41598-017-18079-2. to-villus BMP signalling gradient. Nature Cell Biology 20(8):909e916. https:// [57] Bauer, M.C., O’Connell, D.J., Maj, M., Wagner, L., Cahill, D.J., Linse, S., 2011. doi.org/10.1038/s41556-018-0143-y. Identification of a high-affinity network of secretagogin-binding proteins [68] Sharma, A.K., Khandelwal, R., Sharma, Y., 2019. Veiled potential of sec- involved in vesicle secretion. Molecular BioSystems 7(7):2196e2204. https:// retagogin in diabetes: correlation or coincidence? Trends in Endocrinology doi.org/10.1039/c0mb00349b. and Metabolism 30(4):234e243 https://doi.org/10.1016/j.tem.2019. [58] Malacombe, M., Bader, M.-F., Gasman, S., 2006. Exocytosis in neuroendo- 01.007. crine cells: new tasks for actin. Biochimica et Biophysica Acta (BBA) e Mo- [69] Malenczyk, K., Girach, F., Szodorai, E., Storm, P., Segerstolpe, Å., lecular Cell Research 1763(11):1175e1183. https://doi.org/10.1016/j.bba Tortoriello, G., et al., 2017. A TRPV1-to-secretagogin regulatory Axis controls mcr.2006.09.004. pancreatic b-cell survival by modulating protein turnover. The EMBO Journal [59] Tomas, A., Yermen, B., Min, L., Pessin, J.E., Halban, P.A., 2006. Regulation of 36(14):2107e2125. https://doi.org/10.15252/embj.201695347. pancreatic b-cell insulin secretion by actin cytoskeleton remodelling: role of [70] Hansson, S.F., Zhou, A.-X., Vachet, P., Eriksson, J.W., Pereira, M.J., Skrtic, S., gelsolin and cooperation with the MAPK signalling pathway. Journal of Cell et al., 2018. Secretagogin is increased in plasma from type 2 diabetes patients Science 119(10):2156e2167. https://doi.org/10.1242/jcs.02942. and potentially reflects stress and islet dysfunction. PLoS One 13(4): [60] Li, G., Rungger-Brändle, E., Just, I., Jonas, J.C., Aktories, K., Wollheim, C.B., e0196601. https://doi.org/10.1371/journal.pone.0196601. 1994. Effect of disruption of actin filaments by Clostridium botulinum C2 toxin [71] Gil-Lozano, M., Brubaker, P.L., 2015. Murine GLUTag cells. The impact of food on insulin secretion in HIT-T15 cells and pancreatic islets. Molecular Biology of bioactives on Health: in vitro and ex vivo models. Springer. p. 229e38. https:// the Cell 5(11):1199e1213. https://doi.org/10.1091/mbc.5.11.1199. doi.org/10.1007/978-3-319-16104-4_21. [61] Nevins, A.K., Thurmond, D.C., 2003. Glucose regulates the cortical actin [72] Yagita, K., Okamura, H., 2000. Forskolin induces circadian gene expression of network through modulation of Cdc42 cycling to stimulate insulin secretion. rPer1, rPer2 and dbp in mammalian rat-1 fibroblasts. FEBS Letters 465(1): American Journal of Physiology e Cell Physiology 285(3):C698eC710. https:// 79e82. https://doi.org/10.1016/S0014-5793(99)01724-X. doi.org/10.1152/ajpcell.00093.2003. [73] Bieler, J., Cannavo, R., Gustafson, K., Gobet, C., Gatfield, D., Naef, F., 2014. [62] Lee, J.-J., Yang, S.-Y., Park, J., Ferrell, J.E., Shin, D.-H., Lee, K.-J., 2017. Robust synchronization of coupled circadian and cell cycle oscillators in single Calcium ion induced structural changes promote dimerization of secretagogin, mammalian cells. Molecular Systems Biology 10(7):739. https://doi.org/ which is required for its insulin secretory function. Scientific Reports 7(1): 10.15252/msb.20145218. 6976. https://doi.org/10.1038/s41598-017-07072-4. [74] Drucker, D.J., 2018. Mechanisms of action and therapeutic application of [63] Sanagavarapu, K., Weiffert, T., Ní Mhurchú, N., O’Connell, D., Linse, S., 2016. glucagon-like peptide-1. Cell Metabolism 27(4):740e756. https://doi.org/ Calcium binding and disulfide bonds regulate the stability of secretagogin 10.1016/j.cmet.2018.03.001.